Efficient, Equivariant Predictions of Distributed Charge Models

A machine learning (ML) based equivariant neural network for constructing distributed charge models (DCMs) of arbitrary resolution, DCM-net, is presented. DCMs efficiently and accurately model the anisotropy of the molecular electrostatic potential (ESP) and go beyond the point charge representation used in conventional molecular mechanics (MM) energy functions. This is particularly relevant for capturing the conformational dependence of the ESP (internal polarization) and chemically relevant features such as lone pairs or {\sigma}-holes. Across conformational space, the learned charge positions from DCM-net are stable and continuous. Across the QM9 chemical space, two-charge-per-atom models achieve accuracies comparable to fitted atomic dipoles for previously unseen molecules (0.75 (kcal/mol)/e). Three- and four-charge-per-atom models reach accuracies competitive with atomistic multipole expansions up to quadrupole level (0.55 (kcal/mol)/e). Pronounced improvements of the ESP are found around O and F atoms, both of which are known to feature strongly anisotropic fields, and for aromatic systems. Across the QM9 reference data set, molecular dipole moments improve by 0.1 D compared with fitted monopoles. Transfer learning on dipeptides yields a 0.2 (kcal/mol)/e ESP improvement for unseen samples and a two-fold MAE reduction for molecular dipole moments versus fitted monopoles. Overall, DCM-net offers a fast and physically meaningful approach to generating distributed charge models for running pure ML or mixed ML/MM based molecular simulations. level (0.55 (kcal/mol)/e).

• Publication year

  2026

• Venue

  Unknown venue

• Publication date

  2026-02-06

• Fields of study

  Chemistry, Physics, Computer Science

• Identifiers
  arXiv 2602.07234
• External record

  Open on Semantic Scholar

• Source metadata

  Semantic Scholar

• No claims are published for this paper.

• No concepts are published for this paper.

• QuanDB: a quantum chemical property database towards enhancing 3D molecular representation learning

  2024cited by this paper
• CHARMM-GUI PDB Reader and Manipulator: Covalent Ligand Modeling and Simulation.

  2024cited by this paper
• Completely Multipolar Model for Many-Body Water-Ion and Ion-Ion Interactions.

  2024cited by this paper
• CHARMM at 45: Enhancements in Accessibility, Functionality, and Speed

  2024cited by this paper
• A comprehensive electron wavefunction analysis toolbox for chemists, Multiwfn.

  2024cited by this paper
• Kernel-Based Minimal Distributed Charges: A Conformationally Dependent ESP-Model for Molecular Simulations.

  2024cited by this paper
• Swallowing the Bitter Pill: Simplified Scalable Conformer Generation

  2023cited by this paper
• CHARMM-GUI Membrane Builder: Past, Current, and Future Developments and Applications

  2023cited by this paper
• A Hitchhiker's Guide to Geometric GNNs for 3D Atomic Systems

  2023cited by this paper
• TensorNet: Cartesian Tensor Representations for Efficient Learning of Molecular Potentials

  2023cited by this paper
• Higher-order equivariant neural networks for charge density prediction in materials

  2023cited by this paper
• Systematic improvement of empirical energy functions in the era of machine learning

  2023cited by this paper
• SPICE, A Dataset of Drug-like Molecules and Peptides for Training Machine Learning Potentials

  2022cited by this paper
• Transfer Learning for Affordable and High-Quality Tunneling Splittings from Instanton Calculations.

  2022cited by this paper
• Uncertainty quantification for predictions of atomistic neural networks

  2022cited by this paper
• Molecular Dynamics with Conformationally Dependent, Distributed Charges.

  2022cited by this paper
• MACE: Higher Order Equivariant Message Passing Neural Networks for Fast and Accurate Force Fields

  2022cited by this paper
• The design space of E(3)-equivariant atom-centred interatomic potentials

  2022cited by this paper
• Deep Neural Network Model to Predict the Electrostatic Parameters in the Polarizable Classical Drude Oscillator Force Field.

  2022cited by this paper
• Molecular electrostatic potential analysis: A powerful tool to interpret and predict chemical reactivity

  2022cited by this paper
• Learning Atomic Multipoles: Prediction of the Electrostatic Potential with Equivariant Graph Neural Networks.

  2021cited by this paper
• Equivariant message passing for the prediction of tensorial properties and molecular spectra

  2021cited by this paper
• Impact of the Characteristics of Quantum Chemical Databases on Machine Learning Prediction of Tautomerization Energies.

  2021cited by this paper
• Transfer Learning to CCSD(T): Accurate Anharmonic Frequencies from Machine Learning Models.

  2021cited by this paper
• SE(3)-equivariant prediction of molecular wavefunctions and electronic densities

  2021influential reference
• Cartesian message passing neural networks for directional properties: Fast and transferable atomic multipoles.

  2021cited by this paper
• Geometric and Physical Quantities improve E(3) Equivariant Message Passing

  2021cited by this paper
• The Molpro quantum chemistry package.

  2020cited by this paper
• Machine learning of free energies in chemical compound space using ensemble representations: Reaching experimental uncertainty for solvation.

  2020cited by this paper
• Psi4 1.4: Open-source software for high-throughput quantum chemistry.

  2020cited by this paper
• PhysNet: A Neural Network for Predicting Energies, Forces, Dipole Moments, and Partial Charges.

  2019cited by this paper
• Energy refinement and analysis of structures in the QM9 database via a highly accurate quantum chemical method

  2019cited by this paper
• Visualization of very large high-dimensional data sets as minimum spanning trees

  2019influential reference
• A probabilistic molecular fingerprint for big data settings

  2018cited by this paper
• Automated Topology Builder Version 3.0: Prediction of Solvation Free Enthalpies in Water and Hexane.

  2018cited by this paper
• A reactive, scalable, and transferable model for molecular energies from a neural network approach based on local information.

  2018cited by this paper
• Toolkit for the Construction of Reproducing Kernel-Based Representations of Data: Application to Multidimensional Potential Energy Surfaces

  2017cited by this paper
• The theory of intermolecular forces, 2nd edition, by A. Stone

  2017cited by this paper
• Improved Description of Sulfur Charge Anisotropy in OPLS Force Fields: Model Development and Parameterization.

  2017cited by this paper
• Minimal distributed charges: Multipolar quality at the cost of point charge electrostatics.

  2017influential reference
• Impact of Quadrupolar Electrostatics on Atoms Adjacent to the Sigma-Hole in Condensed-Phase Simulations.

  2016cited by this paper
• Geometric Deep Learning: Going beyond Euclidean data

  2016cited by this paper
• Minimal Basis Iterative Stockholder: Atoms in Molecules for Force-Field Development.

  2016influential reference
• Transferable Atomic Multipole Machine Learning Models for Small Organic Molecules.

  2015cited by this paper
• Distributed Multipoles from a Robust Basis-Space Implementation of the Iterated Stockholder Atoms Procedure.

  2014cited by this paper
• Systematic Improvement of Potential-Derived Atomic Multipoles and Redundancy of the Electrostatic Parameter Space.

  2014cited by this paper
• A Novel, Computationally Efficient Multipolar Model Employing Distributed Charges for Molecular Dynamics Simulations.

  2014cited by this paper
• A supervised fitting approach to force field parametrization with application to the SIBFA polarizable force field

  2014cited by this paper
• Charge Anisotropy: Where Atomic Multipoles Matter Most.

  2014cited by this paper
• Quantum chemistry structures and properties of 134 kilo molecules

  2014cited by this paper
• Leveraging Symmetries of Static Atomic Multipole Electrostatics in Molecular Dynamics Simulations.

  2013cited by this paper
• Scoring multipole electrostatics in condensed-phase atomistic simulations.

  2013cited by this paper
• Atomic multipoles: Electrostatic potential fit, local reference axis systems, and conformational dependence

  2012cited by this paper
• Treatment of Halogen Bonding in the OPLS-AA Force Field; Application to Potent Anti-HIV Agents.

  2012influential reference
• Partial Atomic Charges and Screened Charge Models of the Electrostatic Potential.

  2012cited by this paper
• On the difficulty of training recurrent neural networks

  2012cited by this paper
• Enumeration of 166 Billion Organic Small Molecules in the Chemical Universe Database GDB-17

  2012cited by this paper
• On Extension of the Current Biomolecular Empirical Force Field for the Description of Halogen Bonds.

  2012cited by this paper
• Molecular mechanical study of halogen bonding in drug discovery

  2011cited by this paper
• The electrostatic potential: an overview

  2011cited by this paper
• Current status of the AMOEBA polarizable force field.

  2010cited by this paper
• The role of higher CO-multipole moments in understanding the dynamics of photodissociated carbonmonoxide in myoglobin.

  2008cited by this paper
• Polarizable Force Fields:  History, Test Cases, and Prospects.

  2007cited by this paper
• Beyond the isotropic atom model in crystal structure prediction of rigid molecules: atomic multipoles versus point charges

  2005cited by this paper
• The fundamental nature and role of the electrostatic potential in atoms and molecules

  2002cited by this paper
• Classical Electrodynamics, 3rd ed.

  1999cited by this paper
• Advancing beyond the atom-centered model in additive and nonadditive molecular mechanics

  1997cited by this paper
• Conformational dependence of atomic multipole moments

  1995cited by this paper
• Electrostatic interaction potentials in molecular force fields

  1993cited by this paper
• A well-behaved electrostatic potential-based method using charge restraints for deriving atomic char

  1993cited by this paper
• Cutoff Errors in the Ewald Summation Formulae for Point Charge Systems

  1992cited by this paper
• A TRANSFERABLE DISTRIBUTED MULTIPOLE MODEL FOR THE ELECTROSTATIC INTERACTIONS OF PEPTIDES AND AMIDES

  1990cited by this paper
• Angular Momentum: Understanding Spatial Aspects in Chemistry and Physics

  1988cited by this paper
• An approach to computing electrostatic charges for molecules

  1984cited by this paper
• A NEW FORCE FIELD FOR MOLECULAR MECHANICAL SIMULATION OF NUCLEIC ACIDS AND PROTEINS

  1984cited by this paper
• Quadrupole moments of some simple molecules

  1968cited by this paper

• No citing papers are available for this paper.